Overview

clasp is part of the Potassco project hosted at SourceForge. Source code and pre-compiled binaries are available on the Potassco download page.

clasp is an answer set solver for (extended) normal logic programs. It combines the high-level modeling capacities of answer set programming (ASP) with state-of-the-art techniques from the area of Boolean constraint solving. The primary clasp algorithm relies on conflict-driven nogood learning, a technique that proved very successful for satisfiability checking (SAT).

Unlike other learning ASP solvers, clasp does not rely on legacy software, such as a SAT solver or any other existing ASP solver. Rather, clasp has been genuinely developed for answer set solving based on conflict-driven nogood learning. clasp can be applied as an ASP solver (on SMODELS format, as output by Gringo), as a SAT solver (on a simplified version of DIMACS/CNF format), or as a PB solver (on OPB format).

The following competition results show that clasp is a very versatile solver and suitable for many applications:

• ASP Competition 2009
• 1st in all categories (for problems in the P category, clasp merely printed the unique solution computed by a grounder; unlike with most of the other teams, settings used by the Potassco team were geared towards benchmarks)
• SAT Competition 2009
• 1st in two categories (Crafted SAT+UNSAT and Crafted SAT)
• 2nd in one category (Crafted UNSAT)
• PB Evaluation 2009
• 1st in one category (optimization, small integers, linear constraints: UNSAT answers)
• 3rd in three categories
• ASP System Competition 2007
• 1st in two categories (SCore and SLparse)
• 3rd in one category (MGS)
• PB Evaluation 2007
• 1st in one category (PURE-SAT)

clasp provides different reasoning modes including:

• Enumeration of (Projected) Solutions
• Optimization of Solutions
• Cautious and Brave Reasoning

clasp's search procedure incorporates many advanced features from Boolean constraint solving, among them:

• Conflict Analysis via the FirstUIP Scheme
• Nogood Recording and Deletion
• Backjumping
• Restarts
• Conflict-driven Decision Heuristics
• Progress Saving
• Unit Propagation via Watched Literals
• Dedicated Propagation of Binary and Ternary Nogoods
• Dedicated Propagation of Extended Rules (over Cardinality and Weight Constraints)
• Equivalence Reasoning and Resolution-based Preprocessing

To get an impression of clasp's performance, visit the benchmarking area

For the system and a brief user introduction, visit the resources area

Find the latest information and tools at the Potassco project page